Ten Documents Enterprise Customers Expect You to Have

Most growing companies are in better security shape than their paperwork suggests. Multi-factor authentication is on. Backups run. Laptops are encrypted. Access gets removed when people leave, usually the same week.

Then an enterprise customer asks for the policy that says so, and there isn't one.

This is the single largest category of work in a security review — not fixing controls, but writing down controls that already exist. It is also the cheapest gap to close, because nothing has to be built. It just has to be written, approved, and dated.

Here is the documentation set that enterprise buyers ask for, what each document actually needs to contain, and the three things on the list that are not policies at all.

Why they ask at all

A reviewer on the buyer's side has to record a decision about you that someone else could defend later. A conversation does not survive that. A document does.

So the question behind every request is not "are you secure" but "can you show me something I can put in the file." That framing explains most of what follows, including why a short, honest, current document beats a long, impressive, stale one.

One distinction worth holding onto as you read: a few of these are genuinely mandatory under a standard you might be certified against, and the rest are commercially expected. Buyers ask because their process needs an artifact, not because a law requires one. Where a framework does actually require something, it is noted.

The ten

1. Information security policy. The umbrella document — what you protect, who is accountable, and the principles everything else hangs from. ISO 27001 requires this one outright, under a clause that cannot be scoped out, which is why it appears in nearly every questionnaire. A reviewer checks three things before reading a word of it: is it approved by someone senior, is it dated, and has it been reviewed in the last twelve months.

2. Access control policy. How accounts are created, reviewed, and removed. Where multi-factor authentication is required. How administrative access is granted and how often it is reviewed. The common failure is not the policy but the evidence behind it — most companies can describe their joiner and leaver process perfectly and cannot produce a record of the last access review.

3. Asset inventory. What hardware and software you actually have. The first two CIS Controls (v8.1) are Inventory and Control of Enterprise Assets and Inventory and Control of Software Assets — and the word control is doing real work there. Both expect you to find and remove unauthorized hardware and software, not merely keep a list. For a small company that is a maintained spreadsheet and a periodic look, not a platform.

4. Risk assessment and risk register. A list of the risks you have identified, what you decided to do about each, who owns it, and by when. Reviewers use this to tell the difference between a company that thinks about risk and one that bought a policy pack. An honest register with eight entries and real owners is worth more than a polished one with forty.

5. Incident response plan. Who is called, in what order, what gets contained first, when customers are notified, and who decides. Read your customer contracts before you write the notification section — if an addendum commits you to notifying within 24 or 72 hours, your plan has to match that commitment, and a surprising number do not.

6. Business continuity and disaster recovery. What happens if your primary systems are unavailable, how long recovery takes, and how much data you could lose. Two numbers matter here, and reviewers ask for both: how long until you are back, and how far back your last good copy is. They may call these your recovery time and recovery point objectives. Add the date you last tested a restore. An untested backup is a belief, not a control.

7. Vendor and third-party risk management. How you evaluate the vendors who touch your customers' data, and the list of who they are. The buyer is doing a review of you; they want to know you do something similar downstream. If you process personal data on a customer's behalf, expect to be asked for a sub-processor list naming each one — that requirement usually comes from the data processing agreement rather than from privacy law directly, and standard contractual clauses ask for the name, address and contact for each sub-processor.

8. Secure development practices. If you ship software, how code gets reviewed, how dependencies are tracked and updated, how secrets are kept out of repositories, and how you separate development from production. Worth knowing if ISO 27001 is in your future: secure development is not one control there but a family of them, covering the development lifecycle, security requirements, architecture, coding, testing, outsourced development, and environment separation. A single "secure development policy" rarely covers the ground. Reviewers who are technical will read this one closely and skim the rest.

9. Data classification, retention, and disposal. What categories of data you hold, how long you keep each, and how it is destroyed. Strictly these are three documents sharing one slot, and larger buyers will ask for the retention schedule separately. This is also the area most often missing entirely, and the one privacy-conscious buyers care about most, because it is the difference between "we delete customer data on request" as a sentence and as a procedure.

10. Security awareness training. The programme, and the completion records. The records matter more than the curriculum — a reviewer wants to see that people completed it and when. If you are working toward ISO 27001, note that it separates two things: general awareness for everyone, and documented evidence of competence for the people whose work actually affects security. The second explicitly requires you to retain records. Keep them for both.

Three things on the list that are not policies

Questionnaires mix documents you write with artifacts only someone else can produce. Knowing which is which saves time.

An independent report. A SOC 2 report, an ISO certificate, or a penetration test summary letter. You cannot write these. If you do not have one, say so plainly and say what you do instead — an unanswered question reads worse than an honest "not yet." If you do hold a certificate, send the scope statement with it; that is what tells the reviewer whether the certification covers the product they are buying.

An evidence pack. Screenshots and configuration exports showing the control as it is actually set: the MFA enforcement screen, the backup schedule, the encryption setting. Reviewers consistently ask for exports rather than links, because they need something they can file.

A certificate of cyber-liability insurance. Frequently requested, occasionally contractually required, and it comes from your broker in an afternoon.

Depending on what you sell, a handful of others turn up regularly: encryption and key management, logging and monitoring, backup, change management, and background screening and confidentiality terms for staff. Those usually arrive as questions inside a larger questionnaire rather than as standalone document requests.

If you sell to the federal market, add two more: a system security plan and a plan of action and milestones. NIST SP 800-171 names both. Under Revision 2 — still the version defense contracts run on — they are requirements 3.12.4 and 3.12.2. Revision 3 kept both, moved the system security plan into a new planning family, and added an obligation to protect the plan itself from unauthorized disclosure.

What makes a document credible

The same four things, every time.

  • It is approved and dated, by someone with the authority to approve it.
  • It names an owner. "The security team" is not an owner at a twelve-person company.
  • It has been reviewed within the last year, and says so.
  • It describes what you actually do.

That last one is where template packs fall apart. A downloaded policy set will describe a company with a security operations center, a change advisory board, and quarterly access reviews you have never run. The reviewer reads the policy, then asks a follow-up question, and the answer does not match. Now they have found an inconsistency, and everything else you submitted is in question.

A two-page policy describing what your twelve-person company genuinely does will pass a review that a forty-page borrowed one fails. Write down the real thing, even when the real thing is modest.

The other consistency problem

Your documents have to agree with your questionnaire answers, your contract commitments, and each other. The failures we see most often: an incident response plan promising a notification timeline that contradicts the signed addendum; a retention policy that says ninety days while the privacy notice says twelve months; a questionnaire answer claiming annual access reviews that the policy describes as quarterly.

Nobody catches these by reading one document. They surface when a reviewer reads three, which is exactly what a reviewer does.

Where to start

If you have none of this, do not start with the longest one. Start with the information security policy, the access control policy, and the incident response plan — those three appear in almost every review, and they are the ones a buyer will ask to see first. Then build the evidence folder alongside them, because the second review goes far faster than the first when the artifacts are already collected.

If a questionnaire is already open on your desk, the sorting exercise in what happens after your customer sends a security questionnaire is the faster place to begin.

And if you would rather not write all of this from a blank page, the GCA Security Program Blueprint is the documented program in editable form — the policies on this list, plus the scored assessment, risk register and evidence tracking that go with them.

Writing these is not difficult work, but it is specific work, and it is hard to do convincingly about your own company while a deal is waiting. If that is where you are, book a 30-minute deal-readiness call. If the request arrived as a full questionnaire, Security Questionnaire Rescue starts at $1,500.

Is security holding up a deal?

Book a deal-readiness call